BlazePose: On-device Real-time Body Pose tracking

Code

• github.com/google/mediapipe
  Officialtf★ 34,276
• github.com/geaxgx/depthai_blazepose
  pytorch★ 393
• github.com/VNOpenAI/tf-blazepose
  tf★ 211
• github.com/vietanhdev/tf-blazepose
  tf★ 211
• github.com/jiang-du/BlazePose-tensorflow
  tf★ 112
• github.com/vladmandic/blazepose
  tf★ 19
• github.com/axinc-ai/ailia-models/tree/master/pose_estimation_3d/blazepose-fullbody
  none★ 0

Abstract

We present BlazePose, a lightweight convolutional neural network architecture for human pose estimation that is tailored for real-time inference on mobile devices. During inference, the network produces 33 body keypoints for a single person and runs at over 30 frames per second on a Pixel 2 phone. This makes it particularly suited to real-time use cases like fitness tracking and sign language recognition. Our main contributions include a novel body pose tracking solution and a lightweight body pose estimation neural network that uses both heatmaps and regression to keypoint coordinates.

Tasks

Benchmark Results

Reproductions